226 MR. S. BIDWELL ON CHANGES PRODUCED BY MAGNETISATION IN 



In the case of ring No. 1 (the experiment upon which, as upon all others except 

 ring No. 0, was conducted exactly in the manner described for the iron rod No. 1 

 and the cobalt rod) there was no discrepancy gi-eater than 1 scale division in the 

 two series of 13 observations (columns E and K), except in one instance when there 

 was evidently a mistake. 



The number of pairs of observations with ring No. 2 was 19, in 14 of which the 

 discrepancy did not exceed 1 division. Of the five greater discrepancies all but one 

 occurred with descending currents, when the ring was suffering from the heating effect 

 of the full battery. There is some reason to believe that the demagnetisation was 

 not so perfect in this experiment as in the others. 



Eleven pairs of observations were made with the iron rod No. 2. In two cases 

 there were discrepancies of 2 divisions ; no other discrepancy exceeded 1. 



In consequence of the very great effect produced by small changes of temperature 

 upon the susceptibility of nickel, the pairs of observations with this metal were not 

 nearly so concordant as those with iron and cobalt. The number of double observa- 

 tions made was 11. With 5 of these (including a case in which there was an evident 

 misreading of 5 divisions) the discrepancy did not exceed 1 ; twice it was 2, three 

 times it was 3. One observation of a pair was rejected on account of the evident fact 

 that the demagnetisation had been accidentally omitted.* 



But, though the discrepancies axe great in comparison with those which occurred 

 with the other metals, their absolute magnitude is very small, and it is not probable 

 that any point in the nickel curve, fig. 4, is in error by so much as a millionth of an 

 inch,t or two ten-millionths of the length of the strip. 



ADDENDUM. 



Received June 5, 1888. 



THROUGH the great kindness of Mr. W. H. PBEECE, F.R.S., who allowed me the 

 unrestricted use of the large battery of secondary cells employed in lighting his 

 residence at Wimbledon, I have been able to repeat some of my experiments with 

 much stronger magnetising currents. 



The instrument and the magnetising coil used and the mode of working were 

 exactly the same as before, but it was thought desirable to vary the samples of the 

 metals, in order that it might appear how far the peculiarities already noticed were 

 independent of the particular specimens which had been examined. 



* The demagnetising process involved eight separate operations, including the closing and opening of 

 switches. The omission of any one of these might render the process of no effect. 



f When the discrepancy was 3 scale divisions the probable error of the mean = 0'67 X 1'5 = 1 scale 

 division = 0-000018 mm. = 0'00000072 inch (or less than a millionth) = O'OOOOOOIG of the length (or 

 less than two ten-millionths). 



